GH2 GH2vk Unleashed

cowpunk52 said:
...
However, if you know you want to grade your footage, I would heartily recommend using the GH2vk hack.....
Click to expand...

Thanks for the tip + uploading this stuff! I´ll give it a go tomorrow... grading with the hacked + unhacked footage. And yes, I´m feeling better now :-).
 
Oedipax said:
Hey Sage - by half do you mean GOP=6? Just wanted to make sure. Great looking results, I'm about to try it with 42/6.
Click to expand...

Thank you Oedipax-
Here's a response I gave to Cosimo regarding 6 gop on vimeo:
"Its a balanced hedging of the bet; some argue for 3, others for 12 (the original on 24p). A GOP of 12 includes 1 I frame, 8 B frames, and 3 p frames. By using a GOP of 6, just the middle p frame is replaced w/ an I. This is not a huge loss of efficiency - yet an I frame is a complete refresh of the sensor data reference material, which is great food for the subsequent b and p frames. This means that only 5 frames are suspended between reference material, instead of 11, and a solid p frame is there to bolster the middle, instead of [four] underfed b frames (which are heavily based on what comes before and after). In using 3, there is a more significant loss of efficiency (all 3 p frames must be replaced w/ i, so no p architecture implemented)
In my experience, lower gop tends to have a more organic filmic smooth motion (compare the vimeo's long underfed gop 1080 to the original file)"

Just a little more bitrate is required for one frame out of the twelve in the gop, yet structurally this slight loss of efficiency saves 11 frames from having to survive on the bounty of the last i frame.
 
Can someone post more test footage from the GH2 Hack

Can someone post more test footage from the GH2 Hack

and also throw it in Magic Bullet and throw some cool looks onto it. Someone please get cinematic on the rest of us for our enjoyment! No plants, children playing with toys or boring tests. Challenge yourself to do your test footage like a scene in a movie.
Thats the kinda test footage Id like to see :)
Posting up on Vimeo would be great too.
 
Thanks to everyone for making the effort to test the GH2 hack.
The footage being posted of the 32/42mbps, etc. video looks very clear.

I think it would be helpful if more comparisons to the stock 24mbps could be posted in the same video, for example, contained in one Vimeo video. The thing is that many people, whether they have a camcorder, GH2, etc., end up using Vimeo or Youtube as the final destination for their videos. I do the same thing. But since it's not really, for example, 42mbps on the Vimeo screen, comparisons to the original 24mbps would be helpful to see how much it matters in the end.

I understand that the original file can be made available for download on Vimeo and that some video makers use original quality viewed on their HD monitors, BluRay, HDTVs, etc. to watch high quality footage.
I'm just talking in general terms; that many videos do end up on the internet and nowhere else for distribution.
Thanks again for your work, Vitaly and all.
 
In the statics shots the difference is easy and clear to see for me. the grain / texture in the wall and in the foliage freezes in a constant interval in the unhacked footage. something about 2 times per second. it is annoying because constant freezes in motion hurts the beauty of moving images. In the hacked footage the problem is gone and the grain / texture does not freezes anymore. Also the grain is fine instead of macroblocked. I think this is what people calls organic look. Hacked footage is much more cinematic for me. You must fix your eyes in a single point to view the difference and it is easyer to see in the house wall 0001.mts file.
 
The secret to keep this fine grain organic look in the final deliver media will be to keep the datarate high when encoding to a bluray disc, encode in a 5MBps / 40Mbps and also to use a high bitrate in the renders in the editing workflow. If you work encoding to DCP files to project in a theater the hack will give great results.
 
Yes, It is far better to work with hacked footage, it is a pleasure to edit and look at it in a 1920 x 1080 second monitor while working in the software in the primary monitor. but... we must remember... this difference is negligible when you take into account the SCRIPT, ACTING, and so on...
 
I think the low light situations and high iso will be when/where the hack will be most usefull. The black areas and dark images will take benefits from the high bandwidht to avoid macroblocking / freezing textures. This helps a lot to shot scenes like we can see in the Woody Allen film Midnight in Paris and get a organic film look.

There is a test on vimeo from EOSHD showing night scenes, probably the file for download can show the organic texture at 1920x1080 5MBps encoding.

http://vimeo.com/26611738
 
I set my encoder at 35Megabit (24P). I recorded my drive home from work through the front window, as I figured alot of the details would be rapidly changing.

At some point the camera froze. The image was frozen on the last scene recorded. Power off, nothing. Remove battery and replace, it reboots. Comes right back up.

When I look at the file I see that it is just over the 4GB mark (4.19GB).

The file plays fine and the footage is great. Not sure why it froze up, but if I were using this at an event, I suppose I'd just have to figure a time before the 4GB limit and stop and start very quickly.
 
I did. I ran another test with more detail, and a higher average bitrate (32Mbit), and my camera didn't freeze.

The file was slightly smaller than the last, so I still need to test it further.

Here is file analysis. 35Mbit for 24H, with GOP of 6 and 30 minute time checked.

bitrate.jpg
 
You must log in or register to reply here.
Back
Top